A year ago China’s dexterous-hand sector was a demo-dominated infancy. Capital interest was thin, most makers were still validating concepts, and industry shipments sat at the hundred-to-thousand-unit level, with products suited to trade-show displays rather than stable industrial use or full humanoid integration.
In a single year the picture flipped. Hot money rushed in, new companies formed at speed, and one, Critical Point, reached a US$1 billion unicorn in five months, a record for the sector. According to IT Juzi, full-year 2025 financing in the sector reached 16.88 billion yuan, and the first half of 2026 alone passed 25 billion yuan, more than the whole of last year.
Shipments crossed from thousands to tens of thousands. Estimated 2025 domestic dexterous-hand output exceeded 20,000 units; 2026 market expectations climb to 70,200 units. Technically, the field raced from low degrees of freedom into a high-DOF era across multiple routes. The positioning shifted too, from a robot accessory part toward the core terminal of embodied interaction.
As the bubble clears and capabilities diverge, China’s dexterous-hand industry has formed a clear first tier: four firms on four complementary routes. Inshand follows an industrial-reliability route, Linker Hand a general high-DOF platform, Critical Point a hardware-plus-algorithm manipulation solution, and BrainCo a human-machine bionic fusion route.
Five rulers for a new tier
In the prototype era, any single technical highlight was a core label. At scale, the measure is whether a firm can deliver steadily, create value in real scenes, and build a resilient product matrix. Five criteria capture this: production volume (delivery power), real-scene application (deployment power), route breadth (diversification), applicability (ecosystem fit), and a DOF ladder (product-matrix power).
Inshand is the Toyota of the field: no flashy spec wars, just lean manufacturing, rugged durability and the world’s top delivery volume. A near-decade veteran, it was the first in China to deliver ten-thousand-unit five-finger hands, holds over 60 per cent of the global market, and expects 2026 deliveries of 30,000 to 50,000 units. In 2025, 20 per cent of its overseas hand revenue came from abroad, with more than 2,000 customers across humanoid leaders, manufacturers, universities and medical, commercial and home settings.
Linker Hand is the aggressive one, aiming to become the Nvidia of the embodied era: a platform player that defines the end-effector standard and controls the software-hardware ecosystem. One of few covering linkage, tendon and direct-drive, its matrix spans 11 to 22 degrees of freedom and 3,999 to 99,999 yuan, from research to industry to consumer.
Critical Point is the surprise, perhaps the fastest to a full-stack layout. In its first complete quarter, Q1 2026, it turned operationally profitable. Cumulative deliveries across its range exceed 20,000 units, the OmniHand series over 8,000, and grippers over 10,000. External-customer orders now exceed 40 per cent. Spanning linkage, direct-drive and tendon drive, it ships the OmniHand 3 Lite for 0.8 to 1.3 metre robots and the OmniHand 3 Ultra-M flagship, backed by its DUET dual-layer contact-intelligence architecture.
BrainCo is the only player combining dexterous and bionic hands. Its 2025 Revo 2 was the lightest dexterous hand then on the market at 383 grams, yet lifts 20 kilograms, a 52:1 ratio. This year’s Revo 3 uses direct drive for heavy load and ultra-fine operation, with 70 newtons of grip and 0.01 newton tactile resolution. Total dexterous-hand output nears 10,000 units.
Four routes, one maturing market
The four are not homogenised. One defends industrial volume, one attacks high-DOF, one leverages humanoid synergy, one fuses bionics. Competition is moving through four layers: hardware capability, real physical-operation data, hand-control strategy and embodied models, and ecosystem reach. The winners will embed into open robot platforms. The verdict favours those who close the loop across production, scene, data and algorithm, not those who build the prettiest hand.
Editor’s note: This is an adapted translation of the original OFweek report. It has been trimmed and restructured for readability for an international business audience.